Notes on the Testing of an AEG MOSFET Switch Buffer with Braking
Feb. 3, 2007
Terry Fritz
Circuit Schematic Diagram
Setup
In this test, a small 12V "gel cell" battery and a Pittman gear motor were used.
Picture of Control Circuit
Since this was just for electrical testing, No attempt was made to make it "miniature" to fit inside and AEG.
Voltage (yellow) and Current (blue) across the motor.
The current probe was AC coupled so it does not read sustained DC. However, the startup and breaking currents are clearly shown along with the battery "droop" as the motor comes up to speed. The motor voltage is clearly clamped at just less than 1V by the braking FET Q2 when the switch is released.
Battery Current - Switch "Bounce" and Cross-Conduction Spiking.
As expected, there is a switching region where both FETs are turned on and the battery is effectively shorted. R1 is a very small value to greatly speed through this region to reduce loss. The switch bouncing is actually very well repeated buy the FETs which causes a series of cross conduction spikes. Although the switching could be slowed to "cover" this switch bounce, it would probably increase the losses far more than they are now.
The spikes are current limited to 14 amps by the rather poor transconductance of the IRF5210. With a 12V supply, the cross conduction region is from roughly 5 to 7 volts. There is not enough gate voltage to drive great currents at a low Drain to Source voltage.
IRF5210 Characteristics
High speed, low loss, and current limited spiking like is does no harm and is a "don't care". The poor transconductance of the P-Channel FET is a very good thing here!
Battery Voltage
Gate Voltage
Note that C1 seems to be doing little in this case.
For the next test, the motor was replaced by a 10 amp fast blow fuse for short circuit testing.
Fuse used to short output.
Output voltage and current into a fuse short (before it blew).
Of special note here is the droop at about 30mS! If we look at the gate voltage we see that the battery voltage has dropped out and C1 has discharged to the point where it falls out of saturation.
Q1's gate voltage droop and loss of saturation at 30mS.
In this case, the lead acid battery voltage falls out. The currents would be significantly higher with NiCd or NiMH batteries. However, the very high transconductance of the IRL1404 would still draw the voltage down and Q1 would still loose saturation.
IRL1404 Characteristics
In this state, the power dissipation in Q1 is very high at about 30 amps at say 5 volts for 150W. Assuming the condition is eventually stopped, a small heatsink will protect Q1. Without a heatsink, Q1 will destruct very quickly!
If the battery voltage is forced across the output, the poor transconductance of Q2 again limits the current to about 14 amps. A much better but still worrisome power dissipation.
Normally both FETs run cold. At 13 amps DC, Q1's voltage drop was measured at 0.087 volts for a power dissipation of only 1.131 watts. But that is about the maximum power an un-heatsinked TO-220 FET can withstand. At a higher current, a heatsink is needed. Q2's normal power dissipation is almost "zero".
Summary:
The system works well and as expected. and is pretty hard to destroy.
Q1 should have a heatsink. Both Q1 and Q2 "need" a heatsink to survive a prolonged short to ground or the +battery. The mounting tabs on them are both the FET Drain so they can be electrically connected to a common uninsulated heatsink.
The cross conduction is limited by Q2 and is of no concern.
C1 can ride out a 30mS full loss of voltage and filters the gate voltage very well. A larger cap could be used, but if the dropout lasts longer than 30uS you are probably doomed anyway...
Note that the battery and motor here were "weaker" than in a normal AEG. But there is nothing here to suggest any special problems with a full AEG system.
There is a 5.7M Byte movie here (DIVX) that shows the difference between braked and unbraked operation of the motor.
http://team-titanium.com/~airsoft/AEG-FET-Control/AEG-FET-Brake.AVI
gasma